Connection is Reset during long running handlers, service goes into zombie state

[Ian Cooper]

We get a reset on the connection when the handler exceeds the heartbeat timeout as we don't have any I/O activity on the connection. This is not handled in the consumer receive but in the message pump. which is wrong as it is transport specific and needs handling in the consumer and re-throwing as a channel failure to force re-connection of the consumer. In addition, we need to let clients declare a "long -running handler" if they expect a job to take more than 15s, so that we can fire a connection event on a timer whilst we are processing.